Physical Chemistry Chemical Physics www.rsc.org/pccp

ISSN 1463-9076

Volume 12  |  Number 10  |  14 March 2010  |  Pages 2261–2504

COVER ARTICLE Krylov et al. The effect of π-stacking, H-bonding and electrostatic interactions on the ionization energies of nucleic acid bases

PERSPECTIVE Sheppard and De La Cruz A systematic review of the application of vibrational spectroscopy to the determination of the structures of NO adsorbed on single-crystal metal surfaces

1463-9076(2010)12:10;1-6

PAPER

www.rsc.org/pccp | Physical Chemistry Chemical Physics

The effect of p-stacking, H-bonding, and electrostatic interactions on the ionization energies of nucleic acid bases: adenine–adenine, thymine–thymine and adenine–thymine dimersw Ksenia B. Bravaya,a Oleg Kostko,b Musahid Ahmedb and Anna I. Krylov*a Received 24th September 2009, Accepted 19th November 2009 First published as an Advance Article on the web 6th January 2010 DOI: 10.1039/b919930f A combined theoretical and experimental study of the ionized dimers of thymine and adenine, TT, AA, and AT, is presented. Experimentally observed and computed adiabatic and vertical ionization energies (IEs) for monomers and dimers as well as thresholds for the appearance of the protonated species are reported and analyzed. Non-covalent interactions strongly affect the observed IEs. The magnitude and the nature of the effect is different for different isomers of the dimers. The computations reveal that for TT, the largest changes in vertical IEs (0.4 eV) relative to the monomer occur in asymmetric H-bonded and symmetric p-stacked isomers, whereas in the lowest-energy symmetric H-bonded dimer the shift in IEs is much smaller (0.2 eV). The origin of the shift and the character of the ionized states is different in asymmetric H-bonded and symmetric stacked isomers. In the former, the initial hole is localized on one of the fragments, and the shift is due to the electrostatic stabilization of the positive charge of the ionized fragment by the dipole moment of the neutral fragment. In the latter, the hole is delocalized, and the change in IE is proportional to the overlap of the fragments’ MOs. Relative to TT, the shifts in AA and AT are much smaller due to a less efficient overlap, smaller dipole of A and the large energy gap between ionized states of A and T monomers in the case of AT dimer. The ionization of the H-bonded dimers results in barrierless (or nearly barrierless) proton transfer, whereas the p-stacked dimers relax to structures with the hole stabilized by the delocalization or electrostatic interactions.

I.

Introduction

Oxidative and radiative damage of DNA involves ionization of the individual building blocks, and the nucleic acid bases (NAB) thymine (T), cytosine (C), adenine (A) and guanine (G) are most likely to be ionized. Based on the gas-phase IEs, G is the easiest to ionize, however, the positive charge does not necessarily remain on G and can migrate along the DNA strand as far as 200 A˚1–3 to the other low IE sites. The mechanism of the hole transfer facilitated by thermal fluctuations is not fully understood. Thermal fluctuations induce structural changes in nucleotides and in their local environment that affect so-called site energies, which can be described as IEs of the nucleotides at the specific configuration, and these dynamical changes in the local IEs drive the hole migration. Many factors may affect local IEs of the nucleobases: structural deformations of the nucleobase itself, presence of the backbone, interaction with the solvent, hydrogen bonding, p-stacking, and electrostatics. Thus, understanding how different type of interactions affect the IEs is the first step toward a mechanistic a

Department of Chemistry, University of Southern California, Los Angeles, CA 90089-0482, USA. E-mail: [email protected]; Tel: (213) 740-4929 b Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA w Electronic supplementary information (ESI) available: Supplementary Figures 1–10 and Tables 1–4; geometry data; harmonic frequencies; thermodynamic analysis. See DOI: 10.1039/b919930f

2292 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

understanding of the ionization-induced dynamics in DNA. Another crucial mechanistic question is how does the hole migration occur. Once the site energies of the neighboring bases match, does the hole simply hop from a higher to a lower energy site, or is there a transient hole delocalization over two (or more) bases involved? Similar mechanistic questions about the role of non-covalent interactions in charge transport through molecular wires arise in the context of molecular electronics.4–10 A fundamental understanding of radiation-induced processes in individual bases and their clusters are also important for astrobiology and prebiotic chemistry.11 While IEs of the NAB in the gas phase have been characterized both experimentally12–18 and computationally,19–22 much less is known quantitatively about the effects of different interactions on IEs in a realistic environment. Similar to electronically excited states,23–28 the ionized states are significantly affected by base stacking, base pairing via hydrogen bonding, electrostatic interactions, and solvent effects. The latter have been reported to lower IEs of anionic nucleobases and to reduce the gap between them and phosphate IE suggesting that nucleobase ionization becomes more favorable in water.29,30 Likewise, microsolvation of the nucleobases decreases their IEs by about 0.1 eV per water molecule.16,18 Base pairing and other hydrogen bonding interactions also reduce IEs and oxidation potentials, as discussed in a computational study of Crespo-Herna´ndez et al.31 This journal is

 c

the Owner Societies 2010

Recently, the effects of H-bonding and p-stacking on the IEs of two isomers of the uracil dimer were characterized by high-level electronic structure calculations.32 Earlier studies of the effects of p-stacking on IEs of nucleobases include Hartree–Fock and DFT estimates using Koopmans theorem,33–37 MP2 (Møller–Plesset perturbation theory) and CASPT2 (perturbatively-corrected complete active space self-consistent field) calculations.19,21,38 In this work, we present the combined experimental and theoretical investigation of IEs of AA, TT, and AT dimers in the gas phase. The focus of this work is on the effects of non-covalent interactions, i.e., hydrogen bonding, p-stacking, and electrostatic interactions, on the IEs of the individual nucleobases. We also analyze the changes in electronic structure (i.e., hole localization and proton transfer) induced by these interactions. The structure of the paper is as follows. The next section presents experimental and theoretical details. Section III discusses equilibrium structures and binding energies of the representative neutral dimers, and vertical IEs (VIEs). For selected dimers, we also report optimized geometries of the cations and adiabatic IEs (AIEs). The theoretical results are compared with the experimentally measured spectra in Section IIIF. Our final remarks are given in Section IV.

II. A.

Experimental and computational methods Experimental details

The experiments were performed on a molecular beam apparatus coupled to a 3 meter VUV monochromator on the Chemical Dynamics Beamline at the Advanced Light Source (ALS). The clusters of the NAB are prepared by thermal desorption (T = 495 K) of individual bases (pure beams of A and T), as well as simultaneous desorption of two bases (A and T mixed beam), and cooled in a supersonic beam. In the latter experiments, both homo- and hetero clusters are produced. The thermal vaporization source has been described recently in a publication detailing the microsolvation of DNA bases.18 In the current experiments, the backing pressure of Ar expanded through a 100 mm diameter nozzle was 35 kPa. The clusters are ionized by tunable synchrotron radiation in the 7.4–11.5 eV region, and the ions are detected by a time-offlight mass spectrometer. For each mass, the yield of the ions is measured as a function of the photon energy, which produces photoionization efficiency (PIE) spectra. The typical step size for the PIE scans is 50 meV, and a dwell time of 10 s at a repetition rate of 10 kHz. The differentiation of the PIE curves following the method used by Berkowitz in interpreting photoionization of methanol39 produces a spectrum similar to a photoelectron spectrum from which information about vibrational progressions and other electronic states can be extracted. The differentiation is performed numerically after taking a five points nearest-neighbor average to reduce the effects of noise in the PIE. The accuracy of reported onset energies in the PIE spectra is 0.05 eV. For protonated species (BH+), which arise at mass M + 1, where M is the mass of a neutral monomer, corrections have to be made to account for the natural abundance of isotopes This journal is

 c

the Owner Societies 2010

constituting M. This amounted to 7.47% and 6.46% of the M signal for A and T, respectively, which was subtracted from the M + 1 signal to provide the true BH+ signal. The error in determination of the appearance energies of AH+ and TH+ species is 0.1 eV. The lower accuracy and the small negative signal (see the ESIw) in this case comes from the overlap of the M and M + 1 bands in the mass spectra, in case of which contribution of the isotope-corrected M signal to M + 1 band can be overestimated. B.

Theoretical methods and computational details

Accurate description of equilibrium geometry of nucleobases dimers requires methods capable of treating attractive dispersion forces originating from the correlated motion of electrons. Density functional theory (DFT) often used for geometry optimizations of extended molecules fails to reproduce these interactions.40 The wave function based MP2 method underestimates intermolecular distances,41 and the results are very sensitive to basis set superposition error (BSSE). An inexpensive solution is offered by empirically corrected DFT in which the standard functionals are augmented by R6 terms.42 Geometries of the neutral dimers were optimized with the B3LYP-D dispersion-corrected functional with the s6 scaling factor of 1.05 as suggested by Grimme et al.43 and the 6-31+G(d,p) basis set. The basis set effects as well as the performance of B3LYP-D functional were tested for the AT stacked dimer from the S22 set. We found that increasing the basis set from 6-31+G(d,p) to 6-311++G(2df,2pd) results in a slight increase (0.004 A˚) of the interfragment distance, and that the B3LYP-D/6-31+G(d,p) value differs from the BSSE-corrected RI-MP2/TZVPP result44 by 0.076 A˚. For geometry optimizations of the cationic open-shell species we employed long-range and dispersion-corrected oB97X-D functional45 with the same basis set. This functional includes long-range Hartree–Fock exchange to mitigate the notorious self-interaction error that becomes particularly important in ionized dimers. This functional when used with the 6-311++G(3df,3pd) basis set was shown to describe geometries and binding energies of the weakly bound complexes with the mean absolute errors of 0.064 A˚ and 0.22 kcal mol1, respectively.45 Based on the relatively small basis set dependence of the B3LYP-D equilibrium geometries of the neutral AT dimer, we expect similar errors for the oB97X-D/6-31+G(d,p) structures. The standard SG-1 grid was used for the DFT-D geometry optimization. Tight convergence criteria of 1.2  104, 1  107 and 3  105 for atomic displacement, energy change and gradient, respectively, were used due to the shallow potential energy surface of weakly bound neutral stacked dimers. Binding energies of the neutral and cationic dimers were estimated by oB97X-D/6-311++G(2df,2pd) single point calculations with a fine EML(75,302) grid consisting of 75 points in the Euler–Maclaurin radial grid46 and 302 points in the Lebedev angular grid.47 To verify the structures, the Hessians were computed with the oB97X-D/6-31+G(d,p) method and the EML(75,302) grid for dimer isomers used for ionization energies computations. The oB97X-D/6-31+G(d,p) Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2293

optimized geometries were used for binding energies estimations and Hessian computations. Thermodynamic analysis was performed within rigid rotor–harmonic oscillator–ideal gas approximation (RR-HO-IG) for the standard laboratory conditions (T = 298.18 K, p = 1 atm). The relative populations of different isomers were estimated at standard laboratory conditions, as well as at T = 495 K (experimental desorption temperature). Vertical and adiabatic IEs are computed using the equationof-motion coupled-cluster method with single and double substitutions (EOM-CCSD)48–50 for ionization potentials (EOM-IP-CCSD),51–55 which is capable of describing multiple closely lying electronic states in an accurate, robust and efficient computational scheme. The method is free from artifactual symmetry breaking and spin-contamination that often plagues the description of open-shell species. The EOM-IP-CCSD methodology has been successfully applied to characterize spectroscopic signatures of hole delocalization in ionized non-covalent dimers such as benzene dimers,55,56 H-bonded and p-stacked uracil dimers,32 and water clusters.57,58 VIEs of the monomers and dimers were computed with the EOM-IP-CCSD method and the cc-pVTZ and 6-311+G(d,p) basis sets, respectively. Core electrons were frozen in all calculations. In addition, EOM-IP-CCSD/6-311+G(d,p) calculations also employed resolution of identity (RI) approximation with the auxiliary aug-cc-pVTZ basis set. The errors due to RI are below 1  104 eV for the thymine monomer. In dimer calculations, the virtual orbital subspace for the EOM-IP-CCSD calculations was truncated by using frozen natural orbitals (FNO) approximation59 with the natural population threshold of 99.5%. This approach was tested on thymine and the errors were found be less than 0.1 eV for the six lowest ionized states with respect to the full virtual space results. Even better performance (errors below 0.05 eV) was observed for guanine.59 The errors of the FNO approximation with the same occupation threshold were found to be even smaller for the uracil dimer: the mean error for the ten lowest IEs is 0.03 eV for EOM-IP-CCSD/6-311+G(d,p).59 The lowest AIEs were obtained as the difference between the EOM-IP-CCSD energy of the first ionized state at the cation geometry and the CCSD energy of the reference state at the geometry of the neutral. An energy additivity scheme was used to estimate the EOM-IP-CCSD/cc-pVTZ IEs of the dimers assuming that the basis set effects in the monomers and dimers are similar.32 The approximated EOM-IP-CCSD/cc-pVTZ (large basis, LB) VIEs of the dimers were computed from EOM-IP-CCSD/6-311+G(d,p) (small basis, SB) VIEs using the following expression:32 SB LB SB IELB dimer,i = IEdimer,i + (IEmonomer,j  IEmonomer,j)

(1)

where i and j are the number of the dimer’s ionized state and corresponding ionized state of the monomer. The energy additivity scheme was used for each ionized state of the dimer separately by applying a basis set correction for the corresponding state of the monomer. This can be done only for the states that are of a Koopmans-like character and correlate well with the states of the monomers. When the 2294 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

ionized states acquire a mixed character, manifesting itself in multi-configurational wave functions (e.g., as in the AT stacked dimer), the correction becomes more difficult to apply. Adiabatic IEs were extrapolated using the correction for the lowest vertical IEs. Natural charges for the first ionized state of the dimers were computed with the NBO package60 using the IP-CISD/6-31+G(d,p) electron density. In the IP-CISD method,61 the ionized states are described as linear combination of singly ionized (1h) and ionized and excited (2h1p) determinants derived from the closed-shell Hartree–Fock reference. In this approach, which is an N5 approximation to EOM-IP-CCSD, the HF determinant is employed to describe the reference state instead of the CCSD wave function. Optimized geometries, relevant total energies, and harmonic frequencies are given in the ESI.w All calculations were performed using the Q-CHEM electronic structure program.62

III. A.

Results and discussion Binding energies and structures of the neutral dimers

The NAB dimers form numerous isomers,63–65 and their populations in the molecular beam may be non-thermal. For example, higher energy isomers were observed in several experimental studies using laser desorption.65,66 There are three principally different types of relative fragment arrangement in the NAB dimers, i.e., H-bonded, stacked and T-shaped structures. The T-shaped dimers are predicted to be higher in energy than the H-bonded manifold.67 Moreover, one can expect that intermolecular interactions have the smallest effect on the IEs in the case of T-shaped dimers relative to the H-bonded and stacked dimers. Therefore, we considered only the H-bonded and stacked isomers. Fig. 1 shows the typical structures of the stacked and H-bonded dimers, and defines important structural parameters characterizing relative orientation of the fragments. The values of these parameters for several isomers are given in the ESI.w Representative stacked and H-bonded structures are shown in Fig. 2 and 3, respectively. Binding energies of the several lowest-energy H-bonded and p-stacked AA, TT, and AT isomers are presented in Table 1 and shown in Fig. 2 and 3. For the cations, dissociation energies for B+ + B and several BH+ + (B  H) channels are given. In agreement with numerous previous studies (see, for example, ref. 67–69 and references therein), the most stable dimer is a symmetric (C2h) H-bonded one with two monomers being equivalent (TT-HB1 and AA-HB1). In TT, there are several non-symmetric H-bonded dimers with non-equivalent fragments (3–4 kcal mol1 higher than the lowest-energy one), which are stabilized by electrostatic dipole–dipole interactions (note longer H-bond lengths in TT-HB2 and TT-HB3 relative to TT-HB1). The lowest-energy stacked dimer appears only 8–10 kcal mol1 above the lowest H-bonded one, and is followed by four (TT) or five (AA) more isomers within 2.5 kcal mol1 (see Fig. 2). Close inspection of Fig. 2 reveals important structural differences between stacked TT and AA isomers, which have This journal is

 c

the Owner Societies 2010

Fig. 1 Top: chemical structures of adenine and thymine with atomic numeration used in the text. Bottom: definitions of the main parameters characterizing geometries of stacked and H-bonded NAB dimers. For the stacked dimers, these parameters are: distance between the centers of mass (COM) of the fragments (DCM), tilt angle between the planes of the bases (j), the relative twist of the fragments (y), and the horizontal displacement of the COM (d). For the H-bonded dimers, the most important structural parameters are the distances between the atoms participating in H-bonding (d and d 0 ) and the corresponding angles (a).

important implications for the change of IEs due to the stacking interactions. In TT, the lowest-energy p-stacked isomers have good overlap of the polar NH and CO bonds. The five lowest isomers lying within the 2.5 kcal mol1 range have a favorable overlap of several polar bonds. Thus, TT p-stacked structures are dominated by electrostatic interactions. In AA, the situation is different: in the six lowest isomers (which are within 2 kcal mol1), the fragments are less aligned. Some of the isomers are stabilized by the inter-fragment interactions similar to H-bonding, e.g., five of them have the NH2 hydrogens of the one fragment pointing towards nitrogen atoms of the other fragment. However, the structures reveal that these bonds are relatively long and, consequently, are weaker than regular hydrogen bonds. Thus, they do not appear to be the dominant factor, for example, one of the isomers (AA-ST5) has no H-bonds at all and seems to be stabilized mostly by dipole–dipole interactions and dispersion. Both the TT and AA isomers are floppy: the six lowest frequencies corresponding to relative motions of the fragments vary from 20 to 100 cm1 and are remarkably similar in AA and TT. The upshot of these structural differences is that the fragments in TT structures are aligned better and are more equivalent (in 3 isomers both fragments are identical by symmetry), whereas in AA the fragments have different local environments. Both factors contribute to better orbital overlap and larger hole delocalization in TT, which is confirmed by the shapes of MOs and atomic charges, as discussed below. Binding energies (De and D0), which are presented in Table 1, reveal that ZPE has a small effect on binding energies This journal is

 c

the Owner Societies 2010

Fig. 2 Structures of the TT (top) and AA (bottom) stacked dimers and their binding energies (De, kcal mol1) and selected structural parameters (see Fig. 1) computed at the B3LYP-D/6-31+G(d,p) level of theory.

of the neutrals reducing them by 1–2 kcal mol1 in both H-bonded and stacked dimers. For the AT dimers, we considered two H-bonded dimers (AT-HB1 and AT-HB2) and one stacked (AT-ST1). AT-HB1 is the lowest energy dimer and the AT-HB2 H-bonded isomer is the one observed by IR-UV spectroscopy70 in a molecular beam experiment, which is 2.9 kcal mol1 higher in energy (Table 1). The authors70 reported two structures that could not be distinguished in that experiment. We considered the lowest energy structure of these two. B.

Free energy calculations

In an earlier work on uracil dimer, Hobza and coworkers discussed possible entropy effects, which may favor p-stacked relative to H-bonded isomers.71 DS is negative for dimerization, however, the p-stacked dimers have higher entropy because of floppier structures, and, therefore, DS for their formation is less negative relative to H-bonded ones. A later paper from Hobza and coworkers,67 where they Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2295

Fig. 3 Structures of selected H-bonded dimers of T and A bases. B3LYP-D/6-31+G(d,p) dissociation energies (De, kcal mol1) are given in parentheses.

performed full-dimensional (rather than reduced 6D, as in the uracil study) molecular dynamics (MD) calculations for the microcanonical ensemble, does not show any p-stacked

population in non-methylated pairs except for AA and AT, where they report 3% and 2% of the stacked isomers, respectively. In another MD simulation study from the same authors,68 significant (10, 10 and 25% for AA, AT and TT, respectively) populations of the higher-energy stacked isomers were reported. Thus, first-principle predictions of the populations of different structures are not yet reliable, even under well-defined statistical ensemble conditions. Furthermore, it is not entirely clear how populations in the beam prepared by thermal or laser desorption should be modeled, and whether the evaporation can be considered as an equilibrium process. Supersonic cooling is another non-equilibrium step involved in our experiment.72 Thus, it is not surprising that there is extensive experimental evidence that the populations in these beams are non-thermal (see ref. 72 and references therein). For example, the only AT dimer isomer registered by IR-UV spectroscopy in a molecular beam with gas species formed by thermal evaporation is not the lowest energy one.70 In the case of adenine dimers, the lowest energy symmetric structure was not observed experimentally in the gas phase, while the next in energy structure was identified.73 Thus, the results of our estimates of the thermal populations presented below should not be taken as quantitative predictions of the populations in the experiment. Free energy computations within the RR-HO-IG approximation point to dominant population of H-bonded dimers in molecular beam at 298 and 495 K, the latter corresponds to

Table 1 oB97X-D//oB97X-D/6-31+G(d,p) binding energies (kcal mol1) of several lowest-energy H-bonded and stacked TT, AA and AT dimers De Isomer

State

6-31+G(d,p)

6-311++G(2df,2pd)

ZPE

DZPEa

D0b

TT-HB1

Neutral Cation (HT) Neutral Cation (HT) Neutral Cation Cation (HT) Neutral Cation Neutral Cation Neutral Cation (HT) Neutral Cation Neutral Cation Cation (HT) Neutral Cation Cation (HT) Neutral

19.28 30.27,c 15.56 32.66,c 14.88 28.09 28.80,c 11.49 19.80 10.22 20.07 20.78 34.54,c 10.66 21.92 19.80 21.31 30.28,c 16.89 19.77 31.32,c 12.39

19.33 30.78,c 16.04 32.00,c 15.38 28.46 29.30,c 11.59 19.33 10.46 19.44 19.85 34.06,c 10.55 21.36 19.22 21.14 39.87,c 16.55 19.42 30.90,c 12.38

147.012 145.632 147.104 145.757 146.992 144.635 145.262 146.531 146.052 146.872 145.867 143.715 144.069 144.123 143.738 145.410 144.546 144.861 145.644 144.789 145.180 —

1.056 0.640,c 1.343d 1.148 0.765,c 0.895d 1.036 0.357 0.270, 0.739 0.575 1.060 0.916 0.875 1.167 1.306,c 1. 165d 1.575 0.975 1.158 0.079 0.394,c 0.886d 1.392 0.322 0.713,c 1.204d —

18.22 30.14,c 14.89 31.24,c 14.34 28.82 29.03,c 10.92 18.27 9.55 18.47 18.68 32.75,c 8.97 20.38 18.06 21.06 29.48,c 15.16 19.45 29.70,c —

TT-HB2 TT-HB3 TT-ST1 TT-ST3 AA-HB1 AA-ST1 AT-HB1 AT-HB2 AT-ST1e

39.62d 36.28d 37.44d

35.81d

28.83d 28.31d

39.75d 36.21d 37.26d

34.71d

28.20d 27.89d

39.15d 35.49d 37.03d

33.40d

27.31d 26.69d

a Difference between the ZPEs of the dimer and the monomers. ZPE(A) = 71.274 kcal mol1, ZPE(A+) = 71.489 kcal mol1, ZPE(AH+N1) = 80.098 kcal mol1, ZPE(A–H) = 62.807 kcal mol1, ZPE(T) = 72.978 kcal mol1, ZPE(T+) = 72.014 kcal mol1, ZPE(TH+O9) = 80.624 kcal mol1, ZPE(TH+O9_1) = 80.645 kcal mol1, ZPE(TH+O8) = 80.984 kcal mol1 and ZPE((T–H) ) = 63.878 kcal mol1. ZPE values for the monomers were computed with the oB97X-D/6-31+G(d,p)//oB97X-D/6-31+G(d,p) method and the EML(75,302) grid. b Binding energies corrected by ZPE (D0). c Dissociation to B+ and B. d Direct dissociation to BH+ and (B–H). e The AT-ST1 structure is a transition state at the oB97X-D/6-31+G(d,p) level of theory and a true minimum with no imaginary frequencies optimized with B3LYP-D/6-31+G(d,p). The two geometries are very close.

2296 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

This journal is

 c

the Owner Societies 2010

Table 2 Total H (kcal mol1) and S (cal mol1 K1) in the RR-HO-IG approximation and DH0r (kcal mol1), DS0r (cal mol1 K1) and DG0r (kcal mol1) for the dimer formation from the monomers (T = 298.18 K, p = 1 atm) and relative populations of the isomers at 298 and 495 K

TT-HB1 TT-HB2 TT-HB3 TT-ST1 TT-ST3 AA-HB1 AA-ST1 AT-HB1 AT-HB2

H

S

DH0r a

DS0r

DG0r

pTi

157.851 158.383 158.336 157.614 158.324 154.415 154.698 156.457 156.505

127.915 136.183 137.006 132.060 135.506 129.976 129.895 135.155 134.965

18.433 14.611 13.998 10.930 9.090 18.461 8.878 17.753 15.035

43.575 37.307 36.484 41.430 37.984 40.194 40.275 36.675 36.865

5.44 3.49 3.12 1.42 2.24 6.48 3.13 6.82 4.04

1 3.7 2.0 9.3 2.3 1 9.1 1 9.2

= 298.18 K b

   

102 02 106 106

 108  103

pTi 1 1.4 9.4 9.3 4.1 1 5.7 1 6.0

= 495 K b

   

101 102 104 104

 105  102

1

a

Using De computed at the oB97X-D/6-311++G(2df,2pd)/oB97X-D/6-31+G(d,p) level of theory; H(A) = 76.513 kcal mol , S(A) = 85.085 cal mol1 K1, H(T) = 78.477 kcal mol1, S(T) = 86.745 cal mol1 K1. b The ratio of the populations of the particular isomer relative to the most stable one.

the experimental thermal evaporation conditions (see Table 2). The entropy factor (TDS) is different for the stacked and H-bonded dimers of TT, favoring formation of the stacked dimers. For AA, the entropy contributions for the stacked and H-bonded systems are nearly the same (Table 2). Thus, although the entropy contribution slightly favors p-stacked structures over H-bonded, the overall effect in the differences of the Gibbs free energy is small, i.e., about 1–2 kcal mol1 at 298 K for TT, and even less (0.02–0.06 kcal mol1) for AA. More appropriate treatment of the floppy modes beyond the RR-HO-IG approximation may enhance the effect. To conclude, while it is difficult to estimate populations of different isomers in the beam, there is sufficient experimental and theoretical support for the presence of multiple isomers including, perhaps, even higher-energy stacked ones. C.

Character of the ionized states and VIEs

Fig. 4 compares raw and differentiated PIE curves for the monomers and the dimers from a pure thymine or adenine molecular beam. Also shown are appearance energy curves for protonated adenine and thymine. The protonated species are formed via dissociative photoionization of the respective dimers or larger clusters. The TT spectrum features a prominent red shift (0.35 eV) in the onset (AIE), as well as noticeable changes in the relative intensities of different bands. AA exhibits a strikingly different pattern: the changes in the shape of the first band are smaller and the onset is red-shifted by only 0.10 eV (although, it is difficult to determine the onset due to the low intensity of the first band). Overall, Fig. 4 demonstrates the strong effect of dimerization on the IEs. The changes in IEs due to stacking or H-bonding interactions are determined by orbitals’ overlap and delocalization,33–35 as well as electrostatic perturbations by the local environment. These changes, as well as the fate of the ionized states, can be rationalized within the dimer molecular orbital–linear combination of fragment molecular orbitals (DMO-LCFMO) framework.56,57 Below we discuss VIEs and the character of ionized states of several representative isomers of TT, AA, and AT dimers, and explain the changes due to clustering interactions. Table 3 summarizes vertical and adiabatic IEs of TT, AA, and AT dimers. Fig. 5–7 show the MOs and the corresponding VIEs (as well as the changes in IEs relative to the monomer This journal is

 c

the Owner Societies 2010

values) for the (i) two p-stacked dimers, TT-ST1 and TT-ST3 (Fig. 5); (ii) the lowest-energy symmetric H-bonded dimer, TT-HB1 (Fig. 6); and (iii) two non-symmetric H-bonded dimers, TT-HB2 and TT-HB3 (Fig. 7). These energies are computed by EOM-IP-CCSD/6-311+G(d,p) and extrapolated to EOM-IP-CCSD/cc-pVTZ by using the energy additivity scheme, eqn (1). In the symmetric p-stacked isomer (TT-ST1, D0 = 10.9 kcal mol1), the fragments are equivalent, and the dimer MOs are in-phase and out-of-phase linear combinations with equal weights. The structure, in which the pCC bond of one fragment is on top of the pCC bond of another fragment, results in large overlap of fragments HOMOs. At the equilibrium geometry of the neutral dimer, the ionization from the higher (antibonding) orbital is 0.35 eV lower than that of the monomer, whereas the ionization from the bonding orbital is 0.17 eV higher. The fate of these two ionized states is very different. Ionization from the antibonding orbital increases formal bond order between the fragments, and the fragments move closer to maximize the overlap and increase the bonding interactions (discussed in Section IIIE). Ionization from the lower orbital that has bonding character results in a purely repulsive inter-fragment potential leading to dissociation. Higher states of the cation, which are derived from lower lying MOs, exhibit a similar pattern, with the respective splitting depending strongly on the overlap (for example, IEs corresponding to the ionization from in-plane oxygen lone pairs are much closer to the respective monomer values32). The situation is rather different in a slightly higher TT isomer (TT-ST3, D0 = 9.6 kcal mol1), which has lower symmetry. Because the fragments are no longer equivalent, the dimer MOs are less delocalized and the energy splitting between them is smaller. The corresponding VIEs are much closer to the monomer values, e.g., the lowest VIE is only 0.14 eV lower than that of the monomer. The initial hole is localized, and one may expect that both the 1st and 2nd ionized states relax to a bound ionized dimer. The lowest energy H-bonded dimer (D0 = 18.2 kcal mol1) presents a different pattern. Because the monomers’ HOMOs are of p type, their overlap in the H-bonded structure is small. Consequently, the changes in IE are smaller. Interestingly, both out-of-phase and in-phase orbitals are destabilized, the respective VIEs are 0.18 and 0.06 eV lower than the monomer Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2297

Fig. 4

Raw (A,B) and differentiated (C,D) PIE curves for monomers and dimers of thymine (A,C) and adenine (B,D).

value (Fig. 6). Due to the symmetry constraints, the initial hole is delocalized over the two monomers, however, the corresponding optimized cation structure is not a minimum but a saddle point with a negative frequency along the asymmetric H-transfer mode. Thus, ionization of the H-bonded TT results in barrierless H-transfer forming an asymmetric (TH)+(T–H) dimer in which the charge is localized on the TH fragment, and the unpaired electron on the (T–H) moiety (see Section III E). Quite surprisingly, two higher energy H-bonded isomers (TT-HB2, D0 = 14.9 and TT-HB3, D0 = 14.3 kcal mol1) feature yet another pattern: while orbitals are localized on the monomers, the changes in VIEs are large (0.37–0.40 eV for the 1st ionized state) and comparable to those in the symmetric p-stacked isomer (Fig. 7). In these non-symmetric isomers the fragments are not equivalent, and the changes in VIEs are due to the electrostatic field exerted by the fragments (the dipole moment of the neutral T is 4.11 D), as explained below. The inspection of the EOM amplitudes (given in Fig. 5–7) shows that the ionized states in the TT dimers are of a Koopmans-like character, e.g., the leading amplitudes corresponding to ionization of the respective Hartree–Fock orbitals are above 0.9. The electron correlation, however, is very important for accurate energetics and affects the IEs by more than 1 eV. VIEs for the representative AA dimers are presented in Table 3 and in Fig. 8. The H-bonded isomer (AA-HB1) shows trends similar to TT-HB1, i.e., both 1st and 2nd ionized states are destabilized by 0.12 and 0.09 eV, respectively. The stacked AA dimer behaves differently: the changes in VIEs are smaller than those for TT-ST1 dimer due to the less efficient orbital overlap. Both HOMO and HOMO-1 IEs are reduced by 0.19 and 0.14 eV, respectively. Moreover, some of the higher states show more localized character. 2298 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

Interestingly, two of the dimer states show more substantial configuration interaction, i.e., the leading EOM amplitudes drop to about 0.8. Table 3 and Fig. 9 and 10 present the VIEs for the representative AT dimers. The fragments are non-equivalent owing to their different chemical nature, and, consequently, the MOs are more localized. The lowest VIE for the AT-ST1 dimer (Fig. 9) corresponds to the ionization of A (the NBO charge on the A fragment is 0.99), and is 0.09 eV lower than that of the monomer. The next ionized state is localized on T, and its VIE is 0.24 eV lower than that of bare T, but 0.52 eV above the VIE of A. The H-bonded isomers show a similar trend—the lowest ionization corresponds to the A, and the next one to the T moiety (Fig. 10). An interesting pattern observed in these two isomers is that IEs of the A fragment are affected much less than those of T, at least for the low-lying states. Some of the states in AT dimers are profoundly nonKoopmans, as evidenced by the EOM amplitudes. In these states, Hartree–Fock orbitals do not provide an accurate description of the ionized states, and correlated Dyson orbitals should be employed to characterize the hole.22,74 D. The effect of non-covalent interactions on VIEs: a semi-quantitative analysis To explain the origin of the shifts of the VIEs of the NAB dimers relative to the monomer values as well as splitting between dimer ionized states, and to interpret these quantities in terms of interfragment orbital overlap and electrostatic interactions two simple models were used. The DMOLCFMO approach56,57 can be extended to provide a more This journal is

 c

the Owner Societies 2010

Fig. 5 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for the two stacked TT dimers, TT-ST1 (top) and TT-ST3 (bottom). The energy difference (eV) between the dimer IE and corresponding IE of the monomer and leading EOM amplitudes for the shown orbitals are given in parenthesis.

Fig. 7 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for two non-symmetric H-bonded dimers, TT-HB2 (top) and TT-HB3 (bottom). The energy difference (eV) between the dimer IE and corresponding IE of the monomer and the leading EOM amplitudes for the shown orbitals are given in parenthesis.

quantitative analysis of the changes in the IEs in dimers with equivalent fragments. As discussed in ref. 56, a 3-electronsin-2-orbitals problem can be mapped into a 1-electron-in-2orbitals one. The dimer states are obtained by diagonalizing a 2  2 Hamiltonian matrix of the two interacting monomer states, which yields the following value for the energy splitting, DE12, between the states: DE12 ¼

Fig. 6 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for the lowest-energy H-bonded TT dimer, TT-HB1. The energy difference (eV) between the dimer IE and corresponding IE of the monomer and the leading EOM amplitudes for the shown orbitals are given in parenthesis.

This journal is

 c

the Owner Societies 2010

2ðH11 S12  H12 Þ 1  S212

ð2Þ

where H11 is the energy level of the non-interacting fragments, H12 is the respective coupling matrix element, and S12 is the overlap between the fragments’ MOs. To extract these parameters from the electronic structure calculations, we employ a locally-projected HF procedure75 in which the SCF solution for the dimer is obtained in the basis of the MOs that are strictly localized on the fragments. We use the matrix elements of the Fock operator to parametrize eqn (2), i.e., the F12 matrix element is used as H12, and the diagonal elements are replaced by the orbital energies. These calculations were performed with 6-311+G(d,p) basis set. Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2299

Table 3

Adiabatic and vertical IEs (eV) of the TT, AA, and AT dimers

EOM-IP-CCSD/6-311+G(d,p) State

TT-HB1

TT-HB2

TT-HB3

TT-ST1

TT-ST3

AA-HB1

AA-ST1

AT-HB1

AT-HB2

AT-ST1

1st AIE 1 2 3 4 5 6

8.23 8.88 9.00 10.03 10.09 10.48 10.53

8.03 8.66 9.04 9.77 10.14 10.16 10.29

8.15 8.68 9.10 9.78 9.89 10.17 10.28

8.41 8.71 9.23 10.02 10.07 10.28 10.39

8.34 8.92 9.06 9.91 10.02 10.20 10.43

7.48 8.23 8.26 9.36 9.43 9.49 9.49

7.57 8.16 8.21 9.21 9.25 9.30 9.48

7.57 8.36 8.81 9.57 9.60 9.94 10.31

7.39 8.30 8.64 9.59 9.79 9.81 10.17

— 8.26 8.71 9.61 9.83 9.91 9.99

8.48 8.78 9.30 10.12 10.17 10.39 10.50

8.41 8.99 9.13 10.02 10.13 10.31 10.54

7.50 8.25 8.28 9.39 9.46 9.52 9.51

7.59 8.18 8.23 9.25 9.29 9.31 9.50

7.60 8.39 8.88 9.59 9.64 10.04 10.40

7.42 8.33 8.71 9.62 9.81 9.91 10.26

— 8.28 8.89 9.32 9.51 9.97 10.07

Extrapolated EOM-IP-CCSD/cc-pVTZ (see text) 1st AIE 1 2 3 4 5 6

8.30 8.95 9.07 10.13 10.19 10.59 10.64

8.12 8.73 9.11 9.88 10.25 10.27 10.40

8.23 8.76 9.17 9.89 9.99 10.28 10.39

Fig. 9 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for the stacked AT dimer. The energy difference (eV) between the dimer IE and corresponding IE of the monomer and the leading EOM amplitudes for the shown orbitals are given in parenthesis.

Fig. 8 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for the H-bonded (top) and stacked (bottom) AA dimers. The energy difference (eV) between the dimer IE and corresponding IE of the monomer and the leading EOM amplitudes for the shown orbitals are given in parenthesis.

This scheme was used to estimate splitting energies for the three lowest pairs of ionized states of the symmetric H-bonded AA (AA-HB1) and TT (TT-HB1) dimers, as well as symmetric stacked TT dimer (TT-ST1). To increase the number of states 2300 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

in the analysis, we also included the two H-bonded and one stacked symmetric cytosine dimers (see the ESIw). Overall, 18 pairs of ionized states were included in the analysis. Fig. 11 (upper panel) compares the splittings computed by EOM-IP-CCSD/6-311+G(d,p) with the values estimated by eqn (2). The two sets of DE12 show good correlation (R2 = 0.96), except for the three points marked by the arrows that correspond to the ionized states of a non-Koopmans character (i.e., having large EOM amplitudes for more than one orbital). Since eqn (2) is derived based on a one-electron model, these states were excluded from the fitting procedure. Fig. 11 (lower panel) also shows that the splittings correlate well (R2 = 0.91) with orbital overlap S12. Thus, the magnitude of splittings in the symmetric dimers can be well reproduced by a simple one-electron model parametrized by the locally-projected HF calculations and is mainly determined by orbital overlap. The origin of large shifts in IEs in non-symmetric H-bonded dimers (TT-HB2 and TT-HB3) is different. We analyzed this This journal is

 c

the Owner Societies 2010

Fig. 11 Energy splittings between the ionized states computed by EOM-IP-CCSD/6-311+G(d,p) versus the DMO-LCFMO values (upper panel) and the orbital overlap (lower panel). Fig. 10 The MOs and respective VIEs (eV, EOM-IP-CCSD/ 6-311+G(d,p) extrapolated to cc-pVTZ) for the H-bonded AT-HB1 (top) and AT-HB2 (bottom) dimers. The energy difference (eV) between the dimer IE and corresponding IE of the monomer and the leading EOM amplitudes for the shown orbitals are given in parenthesis.

effect by computing the energy of electrostatic interactions between the atomic NBO charges on the ionized fragment with the dipole moment of the other fragment using the following expression: E¼

X

Eðri ; yÞ " #   X qi jDj 1 j Dj 2 3 ¼ cos y þ ð5 cos y  3 cos yÞ þ . . . 8 ri r2i i i

ð3Þ where ri is the distance between atom i of fragment 1 and COM of fragment 2, qi is a NBO charge on the ith atom, |D| is the norm of the dipole moment vector, and y is the angle between dipole moment vector and ri. The center of the dipole moment is positioned in the COM of fragment 2. Only the first two terms of the expansion were taken into account. The partial charges and the dipole moment were computed for the isolated ionized and neutral fragments at their geometries in the dimer. This model is only valid for the non-symmetric dimers with the localized ionized states. The computed energies for the eight states of the two asymmetric H-bonded This journal is

 c

the Owner Societies 2010

Fig. 12 Dimer VIE shifts relative to the monomer versus charge– dipole interaction energy for eight states of the asymmetric TT-HB2 and TT-HB3 dimers and two states of the asymmetric CC dimer, see text.

TT dimers and two states of cytosine dimer (see the ESIw) correlate well with the observed shifts in VIE, as shown in Fig. 12. Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2301

E. Equilibrium structures of the ionized dimers and their adiabatic IEs Changes in geometries of symmetric stacked isomers upon ionization (for details see the ESIw) are consistent with DMO-LCFMO predictions—the fragments adjust to maximize the orbital overlap. Because the orbitals are not completely delocalized, the relaxation does not necessarily reduce the DCM. For example, in TT-ST1 the distance between COM actually increases, however, the fragments also become more parallel (tilt angle changes from 81 to 01), and the horizontal displacement decreases. TT-ST1 cation relaxes to the C1 symmetry structure. The TT-ST1 structure optimized with C2 symmetry constraints lies slightly higher in energy (0.4 kcal mol1) and is characterized by two imaginary frequencies, corresponding to the tilt motion and relative shift of the two thymine rings. These changes reduce the distance between two double CC bonds (hosting most of the HOMO density). The optimized structure of TT-ST3 isomer (that has a localized hole) shows 0.4 A˚ increase in DCM and also a large increase in horizontal displacement (1.22 A˚). This structure is stabilized by the CO group of the ‘‘neutral’’ fragment pointing towards the positive charge localized on the CQC bond of the ‘‘ionized’’ fragment. The partial covalent bond formed between the fragments due to ionization results in an increase in binding energies relative to the neutrals. For example, D0 of TT-ST1 increases from 10.92 kcal mol1 (neutral) to 18.27 kcal mol1 (cation), see Table 1. The lowest AIE of TT-ST1 is 8.48 eV. The optimization of the non-symmetric stacked dimer, TT-ST3, converges to a structure in which the charge is localized on one fragment: the NBO charges for the optimized cation are 0.85 and 0.15 for the two T fragments. D0 for this dimer cation is 18.47 kcal mol1, which means that this isomer is slightly below the symmetric isomer with the delocalized hole. The respective adiabatic IE is 8.41 eV. Energetically, the relaxation in the stacked dimers is larger relative to the monomers, i.e., the VIE-AIE difference in T is 0.23 eV,76 whereas in TT-ST1 and TT-ST3 it is 0.3 and 0.58 eV, respectively. The relaxation in A is 0.25 eV, to be compared with 0.59 in AA-ST1. Ionization of H-bonded dimers is followed by a barrierless or nearly barrierless interfragment proton transfer for most of the considered systems. Different H-bonding patterns are likely to lead to different dissociation products. The energy diagram for the neutrals, cations and dissociation products for several dissociation channels is shown in Fig. 13. The ionization of symmetric H-bonded TT (TT-HB1) results in barrierless H-transfer forming an asymmetric (TH)+(T–H) dimer in which the charge is localized on the TH fragment (the NBO charge for the TH+ moiety is 0.86). The relaxation energy of the cation relative to the neutral’s geometry is 0.65 eV (Table 3). The optimized symmetric structure, which is a saddle point with one imaginary frequency, is 0.11 eV below the vertical geometry (oB97X-D/6-311++G(2df,2pd)//oB97X-D/ 6-31+G(d,p), not shown). The adiabatic IE is rather low and equals 8.30 eV. The dissociation energy of this isomer is 1.3 eV (30.1 kcal mol1) to T and T+ , whereas the dissociation energy producing TH+(O9) and (T–H) is 1.7 eV (Fig. 13). 2302 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

Fig. 13 Relative energies of the ionized dimers and the dissociation products with respect to the lowest energy neutral dimer for each system (TT, AA and AT). The energies are computed with oB97X-D/ 6-311++G(2df,2dp)//oB97X-D/6-31+G(d,p) and include ZPEs. The arrows indicate the channels for direct dissociation of the cationic H-transferred h-boned dimers.

TT-HB2 also undergoes barrierless proton transfer resulting in 0.61 eV relaxation (Table 3). The respective AIE is 8.12 eV, and dissociation energies to T and T+ and TH+(O8) and (T–H) are 31.2 and 35.5 kcal mol1. The true minimum on the cation potential energy surface for TT-HB3 was located for the structure without proton transfer. However, one of the hydrogen bond is already very short in this structure (1.392 A˚) and, hence, one can expect a low barrier and fast interfragment proton transfer in this case as well. The structure with proton transferred from one base to another for the TT-HB3 lies below this minimum by only 1 kcal mol1. The overall relaxation of the cation from vertical to the H-transferred structure is 0.53 eV (Table 3). Dissociation energies to T and T+ and TH+(O9_1) and (T–H) are 29.0 and 37.0 kcal mol1 for the TT-HB3 dimer. The lowest stacked AA isomer converges to a structure that has smaller DCM and j, which increases the orbital overlap with an AIE of 7.59 eV. AA-HB1 converges to a symmetric structure with all frequencies being real, however, this minimum is very shallow and may not be able to support a single vibrational state, e.g., an unrelaxed scan along the H-transfer mode (that has an unphysically large frequency of 5 812 cm1) shows a barrier of only 2 kcal mol1. The proton transfer for the adenine dimer cation has recently been characterized by both theory and experiment.77,78 Although the transition state for this process was successfully located, the inclusion of ZPE makes this process barrierless.77 The reported energy of the proton-transferred structure relative to the vertically ionized state calculated with B3LYP/6-31+G** is 6.9 kcal mol1 (without ZPE correction), which differs from our values computed with more reliable methods, i.e., oB97X-D/6-311++G(2df,2pd) and EOM-IP-CCSD/6-311+G(d,p) yielding VIE-AIE energy This journal is

 c

the Owner Societies 2010

difference of 19 kcal mol1 and 17 kcal mol1, respectively. The agreement between these two estimates points to the problematic behavior of B3LYP in the case of open-shell species. To summarize, the H-bonded TT and AA dimers show barrierless (or almost barrierless) proton transfer forming asymmetric [e.g., (BH)+(B–H) ] structures accompanied by large energy relaxation (about 0.6 eV). These results are in agreement with previous studies suggesting increased acidity of NAB upon ionization,36,79–84 and demonstrate that H-transfer in the ionized states is even more efficient than in the electronically excited states, where the reduced barriers have also been reported.85,86 It has been suggested that coupling of the proton transfer to the hole hopping process slows down the charge mobility in a DNA double strand by separating the charge and the unpaired electron.80 The AT-HB1 dimer presents an interesting case. The hole is initially localized on the adenine base and one can expect proton transfer from A to T due to the increased acidity of adenine upon ionization. Indeed, relaxation of the cation leads to a structure with a very short hydrogen bond (1.485 vs. 1.801 A˚ for neutral) in which adenine is a proton donor. The length of the second hydrogen bond is increased from 1.840 to 2.197 A˚ upon ionization. The proton transfer reaction from adenine to thymine for the AT Watson–Crick base pair has been previously studied using B3LYP/6-31G(d,p) approach.84 The reported transition state and product energies relative to the cation minimum structure without proton transfer were 1.64 kcal mol1 and 1.22 kcal mol1, respectively. However, according to our results, the structure resulting from hydrogen transfer from thymine to adenine lies much lower in energy (0.79 eV below the vertically ionized state, see Table 3). The barrier for the proton transfer reaction from T to A can be significantly higher than in the case of A to T proton transfer due to the longer H-bond. AT-HB2 shows very similar trends, see Table 1. Geometry optimization of the cationic AT-ST1 dimer results in a hydrogen-bonded structure. This can be a result of a shallow potential energy surface of the dimer cation with respect to the change of the tilt angle or an artifact of DFT-D applied to the case the molecule with two competing type of interactions: stabilization caused by orbital overlap (partial covalent character of the dimer cation) and stabilization of the cation through strong hydrogen bonding. F.

much smaller population of larger clusters, these contributions should not be large. Examination of a mass spectrum at 10 eV (see the ESIw) reveals that under our experimental conditions, trimer contributions are around 1% to the total signal, while approximately 30% of the signal correspond to the dimer and protonated monomer, and 69% of the signal arises from the monomer. The protonated monomer arises from dissociative photoionization of the dimer and the respective branching ratios to BH+ and B2+ where B is the molecule of interest depend upon photon energy as can be seen in the different shapes of the PIE curves shown in Fig. 4. Hence we report a combined percentage for the protonated monomer and dimer. Fragmentation of the dimer giving rise to B+ + B contributes to the monomer signal. An examination of the first band of thymine monomer in Fig. 4C shows a broad structured peak extending from 8.7 to 9.6 eV. The signal in this region could also arise from fragmentation of the dimer, however it is difficult to separate the two contributions in our derived PIE. Our calculations for thymine show (Table 4) that this channel opens up around 9.2 to 9.5 eV (depending on the isomer) suggesting that this is indeed occurring.

Experimental PIE curves and theoretical IEs

Fig. 14 shows differentiated PIE spectra for AA, TT for the case of a pure beam of each base and a differentiated PIE of AT dimer for the mixed beam of A and T, as well as computed VIEs and the lowest AIEs for several isomers. Also indicated are threshold energies for the dissociation of the dimers to the protonated and ionized monomers. The onset of the curve corresponds to the lowest AIE, whereas the maxima can be interpreted as VIEs. One important feature of the differentiated PIE curve is that we cannot exclude contributions from fragments of higher clusters, which are likely to be produced via ionization to repulsive states, especially at higher energies. This has been shown to occur in the case of thymine and adenine by Stolow and coworkers.27 However, due to This journal is

 c

the Owner Societies 2010

Fig. 14 Differentiated PIE curves and calculated vertical and adiabatic IEs for the selected isomers. The thresholds for the H-bonded dimers dissociation to B+ and B as well as BH+ and (B–H) are indicated by black bars. Positions of ionized states of different isomers are differentiated by color and height according to the order given in the legends.

Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2303

Table 4 Energy thresholds (eV) for the appearance of BH+ and B+ formed by the dissociation of ionized dimers computed with oB97X-D/ 6-311++G(2df,2pd)//oB97X-D/6-31+G(d,p) and including ZPE Species

Channel

TT-HB1 9.91 TT-HB3 TH+(O9_1) 9.71 TT-HB2 9.54 TH+(O8) TT-ST3 9.17 T+ AT-HB1 8.82 AH+ AT-HB2 + 8.63 AH (N1) AA-ST1 8.51 A+

TH+(O9)

a

AT-HB1 10.03

AT-HB2 10.18a

TT-ST1 TT-HB3 9.22 9.38 AA-HB1 8.97 AT-ST1 8.61a

TT-HB2 9.41

TT-HB1 9.54

AT-HB1 2 AT-HB1 AA-HB1 8.78 8.91 8.94

Without ZPE correction.

Furthermore, the PIE spectra of the dimer will exclude the contributions from the dissociative states, e.g., higher state of the symmetric dimers corresponding to the ionization from the bonding orbitals. Similar difficulties arise in other types of experiments involving intermediate states, see, for example, the discussion of the electron-ion coincidence spectra in ref. 27. As follows from Fig. 14 and Table 3, the AIEs of both TT-ST1 and TT-ST3 are very close to the experimental onset (8.45 eV). AIEs for TT-HB1, TT-HB2 and TT-HB3 dimers fall into the lower energy region. However, due to significant geometrical relaxation in the H-bonded isomers, the Franck–Condon factors (FCFs) are expected to be small, so it is not clear if these isomers can be seen at the onset. The sharp rise of the experimental curve suggests more favorable FCFs, as in, for example, stacked TT dimers. As follows from the calculations, the first band can contain contributions from the states derived from HOMO and HOMO-1 ionization of T, whereas the second band can contain four more ionized states. A large red shift in vertical position is consistent with the computed values of the H-bonded isomer, whereas the lowest VIE of TT-ST1 and H-bonded TT-HB3 appear to coincide with a lower energy shoulder. Since every second line of the symmetric dimers will not appear in the dimer PIE curves (because of their dissociative nature), the large red shift is not accompanied by a broadening of the lines. The second band presents clear evidence of at least two types of isomers: the lowest-energy H-bonded isomer explains lower and higher energy features, whereas the band maximum corresponds to the ionization derived from either one or two p-stacked isomers, or higher-energy asymmetric H-bonded TT-HB2 and TT-HB3. The change in the relative intensities of the two bands can be explained by the fact that the cation states giving rise to the second band are energetically above the dissociation threshold (for the lowest electronic states), and are, therefore, predissociative. The appearance thresholds for the protonated species are also consistent with the computed values, as discussed below. 2304 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

The AA curve has a different shape: it contains a small first band extending from 8.1 to 9.1 eV, followed by three very broad bands. Dimerization affects the two lowest ionized states of AA less than in the case of TT, which can be explained by more efficient orbital overlap in the latter due to the better aligned fragments (in p-stacked isomers) and more localized monomers MOs. The dipole moment of A is also smaller, i.e., 2.61 D in comparison to 4.11 D in T (RI-MP2/cc-pVTZ), which results in smaller shifts in asymmetric H-bonded structures. The computed VIEs of AA show the changes from the monomer values of 0.12 and 0.19 eV for the lowest ionized state of H-bonded and stacked dimers, respectively, and of 0.08–0.21 eV and 0.08–0.29 eV for the upper states (Fig. 8). The AIE value for proton-transferred AA-HB1 is 7.50 eV, which is 0.60 eV below the onset of the PIE curve, 8.10 eV, whereas the symmetric H-bonded structure corresponds to the AIE of 8.15 eV. The AIE for AA-ST1 is 7.59 eV. Large geometric relaxation and poor FCFs may be responsible for the low signal at the onset. Differentiated PIE spectrum of the AT dimer consists of one weak band extending from 8.0 to 9.2 eV with the onset of 8.05 eV and the maximum at 8.8 eV, followed by second unstructured and steeply rising band. Calculated VIEs and AIEs of the H-bonded and p-stacked isomers show that the low energy shoulder of the first band corresponds to the states derived from ionization of A, whereas the band maximum can be assigned to the states derived by ionization of T. The second band (rising almost linearly) covers at least seven ionized states of both AT-HB1 and AT-HB2. Overall, the broad structure of the AT derived PIE spectrum does not allow us to differentiate between contributions from ionization of different isomers to the total intensity. This spectra does not show any clear evidence of the presence of p-stacked isomers. Fig. 14 also shows the computed energy thresholds for the appearance of the protonated species, TH+ and AH+ (summarized in Table 4 and Fig. 13). Note that different isomers of TH+ are considered for TT-HB1, TT-HB2, and TT-HB3 dissociation, as well as different AH+ isomers for the AT-HB1 and AT-HB2 dimers. As is evident from Fig. 13 and the corresponding D0 (see Table 1), the dissociation to the H-transferred fragments is energetically unfavorable relative to the dissociation to B+ and B in TT and AA. In TT H-bonded dimers, the difference between the two channels varies from 4 to 9 kcal mol1 for different isomers, however, in AA they are less than 1 kcal mol1 apart. In AT, the AH+ + (T–H) channel becomes energetically favorable by about 2–3 kcal mol1 relative to A+ and T. Overall, the BH+ + (B–H) channel appears to be very efficient in all species, as follows from the large yield of the protonated species in this study, as well as in ref. 27, 78 and 87. This is not surprising in view of barrierless H-transfer observed in this work for two TT H-bonded isomers and by Park et al.77 for the AAHB1 dimer. The experimental onsets for TH+ and AH+ for the pure molecular beams are 9.20 eV and 9.10 eV, respectively. The yield of TH+ slowly rises for the photon energies between 9.20 and 9.50 eV. Above 9.5 eV there is strong increase in the yield of TH+, especially prominent for 9.7–9.9 eV photon energies. The TH+ onset for the mixed A and T molecular This journal is

 c

the Owner Societies 2010

beam does not change, however, the AH+ appearance energy shifts to 8.90 eV (see the ESIw). Neglecting contributions from larger clusters, AH+ and TH+ can be produced from homo- and hetero-dimers. The computed thresholds can be compared to the experimental onsets assuming that the dissociation does not involve a barrier. For the lowest energy isomer, TT-HB1, the TH+ + (T–H) channel becomes energetically accessible at 9.91 eV, which is 0.4 eV above the sharp rise in TH+ signal (9.5 eV). The low onset energy value can only be explained by the presence of a higher-energy isomer. The most likely candidates are p-stacked TT dimers, for which these channels open up at 9.3–9.6 eV for different TH+ products, assuming barrierless proton transfer (see Fig. 13). Dissociation of the more populated TT-HB2 and TT-HB3 dimers could explain the sharp increase in the signal of TH+ species at 9.5–9.7 eV. The most populated TT-HB1 dimer gives rise to the TH+ signal for the photon energies above 9.9 eV. Note, however, that a possible barrier along the dissociation coordinate would result in a higher-energy onset. The threshold for the TH+ appearance from AT-HB1 and AT-HB2 is much higher, i.e., 10.03 and 10.18 eV, respectively. The experimentally observed onsets for AH+ for pure (9.10 eV) and mixed beams (8.90 eV) are in relatively good agreement with 8.97 and 8.82 eV values calculated for the AA-HB1 and AT-HB1 dimers, respectively. The threshold energy for dissociation of the lowest stacked AA and AT isomers is considerably lower (Fig. 13). Thus, both differentiated PIE and the appearance thresholds of the protonated species support the presence of more than one lowest energy H-bonded dimer and are also consistent with the presence of TT stacked dimers. However, experimental data show no evidence for the presence of stacked AA or AT isomers.

IV.

Conclusions

We present a combined theoretical and experimental study of the ionized dimers of NAB thymine and adenine, TT, AA, and AT. AIEs of TT, AA and AT as well as appearance energies for AH+ and TH+ species are experimentally measured for the first time. The corresponding onset energies for the PIE spectra are 8.45  0.05 eV, 8.10  0.05 eV and 8.05  0.05 eV for the TT, AA and AT dimers, respectively. Appearance energies of AH+ and TH+ are 9.10  0.1 and 9.20  0.1 eV for pure adenine and thymine molecular beams, respectively. Non-covalent interactions strongly affect vertical and adiabatic IEs. In TT, the largest changes in VIEs between monomer and dimer (0.4 eV) were observed in the asymmetric H-bonded and symmetric p-stacked isomers, whereas in the lowest-energy symmetric H-bonded cluster the shift is much smaller (0.1 eV). The origin of the shift and the character of the ionized states is different in asymmetric H-bonded and stacked isomers. In the former case, the initial hole is localized on one of the fragments, and the shift is due to the electrostatic stabilization of the positive charge of the ionized fragment by the dipole moment of the neutral fragment. In the latter, the hole is delocalized, and the change in IE is proportional to overlap of the fragments’ MOs. The fate of the ionized states This journal is

 c

the Owner Societies 2010

is also different: the H-bonded ones undergo barrierless (or almost barrierless) proton transfer, whereas the symmetric p-stacked ones form structures with a partially covalent bond between the fragments. For all the isomers, ionization increases the binding energy by about 10–20 kcal mol1. The onset of the PIE curve of TT dimer is at 8.45  0.05 eV, which agrees well with that of p-stacked isomers. Large geometric relaxation upon ionization (H-transfer) in the H-bonded dimers suggests poor FCFs, whereas the shape of the experimental curve is more consistent with the p-stacked isomers for which the relaxation is less. The differentiated PIE curves show evidence of several isomers, especially in the second band derived from the ionization from lower orbitals. The sharp rise in the signal of TH+ at 9.50–9.70 eV agrees well with the calculated dissociation threshold energies of the asymmetric H-bonded dimers, whereas the threshold for the lowest energy H-bonded isomer is considerably higher (9.91 eV). The AA dimer behaves differently. The shifts in IEs are much smaller due to more delocalized orbitals and less polar character of the adenine monomer (e.g., the dipole moment of A is almost half that of T). The latter results in less aligned p-stacked structures characterized by less efficient overlap between fragment molecular orbitals. A more delocalized orbital of A results in larger geometrical relaxation and poor FCFs for the lowest band in A and, consequently, AA. The onset of the PIE curve for AA is 8.10  0.05 eV, however, the computed AIE for the lowest isomer (proton transferred H-bonded) is much lower (7.50 eV). Additional geometric relaxation due to proton transfer is likely to be responsible for the unfavorable FCFs. The threshold for AH+ appearance is 9.10  0.1 eV, which is in reasonable agreement with the computed threshold for the lowest energy AA isomer (symmetric H-bonded). The measured onset of the PIE curve for the AT dimers is 8.05  0.05 eV. The analysis of VIEs reveals that the states derived from thymine ionization exhibit larger shifts that those derived from ionization of A. Some of the AT states feature profoundly non-Koopmans character. The pattern that emerges from our study is that the IE of T is affected more by non-covalent interactions (both stacking and H-bonding) than those of A (which has lower IE). Thus, the differences in IEs between T and A decrease due to these interactions, which might play a role facilitating efficient charge transfer through the DNA strands.

Acknowledgements This work is conducted under auspices of the iOpenShell Center for Computational Studies of Electronic Structure and Spectroscopy of Open-Shell and Electronically Excited Species supported by the National Science Foundation through the CRIF:CRF CHE-0625419+0624602+0625237 grant. O. K. and M. A. acknowledge support by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division of the US Department of Energy under contract No. DE-AC02-05CH11231. We are grateful to Vadim Mozhayskiy for creation of the cover art accompanying this paper. Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2305

References 1 M. E. Na´cuteun˜ez, D. B. Hall and J. K. Barton, Chem. Biol., 1999, 6, 85. 2 P. T. Henderson, D. Jones, G. Hampikian, Y. Kan and G. B. Schuster, Proc. Natl. Acad. Sci. U. S. A., 1999, 96, 8353. 3 F. D. Lewis, R. L. Letsinger and M. R. Wasielewski, Acc. Chem. Res., 2001, 34, 159. 4 G. R. Hutchison, M. A. Ratner and T. J. Marks, J. Am. Chem. Soc., 2005, 127, 16866. 5 M. Taniguchi and T. Kawai, Phys. E., 2006, 33, 1. 6 C. Nogues, S. R. Cohen, S. Danube, N. Apter and R. Naaman, J. Phys. Chem. B, 2006, 110, 8910. 7 J. R. Heath and M. A. Ratner, Phys. Today, 2003, 56, 43. 8 F. C. Grozema, S. Tonzani, Y. A. Berlin, G. C. Schatz, L. D. A. Ratner and M. A. Siebbeles, J. Am. Chem. Soc., 2008, 130, 5157. 9 R. H. Goldsmith, O. DeLeon, T. M. Wilson, D. FilkensteinShapiro, M. A. Ratner and M. R. Wasilewski, J. Phys. Chem. A, 2008, 112, 4410. 10 F. C. Grozema and L. D. A. Siebbeles, Int. Rev. Phys. Chem., 2008, 27, 87. 11 M. K. Shukla and J. Leszczynski, in Radiation Induced Molecular Phenomena in Nucleic Acids, ed. M. K. Shukla and J. Leszczynski, Springer, Berlin, 2008, p. 1. 12 D. Dougherty, K. Wittel, J. Meeks and S. P. McGlynn, J. Am. Chem. Soc., 1976, 98, 3815. 13 S. Urano, X. Yang and P. R. LeBrenton, J. Mol. Struct., 1989, 214, 315. 14 G. Lauer, W. Scha¨fer and A. Schweig, Tetrahedron Lett., 1975, 16, 3939. 15 C. Yu, T. J. O’Donnel and P. R. LeBreton, J. Phys. Chem., 1981, 85, 3851. 16 S. K. Kim, W. Lee and D. R. Herschbach, J. Phys. Chem., 1996, 100, 7933. 17 H. Satzger, D. Townsend and A. Stolow, Chem. Phys. Lett., 2006, 430, 144. 18 L. Belau, K. R. Wilson, S. R. Leone and M. Ahmed, J. Phys. Chem. A, 2007, 111, 7562. 19 E. Caue¨t, D. Dehareng and J. Lie´vin, J. Phys. Chem. A, 2006, 110, 9200. 20 D. Roca-Sanjua´n, M. Rubio, M. Mercha´n and L. Serrano-Andre´s, J. Chem. Phys., 2006, 125, 084302. 21 E. Caue¨t and J. Lie´vin, Adv. Quantum Chem., 2007, 52, 121. 22 H. R. Hudock, B. G. Levine, A. L. Thompson, H. Satzger, D. Townsend, N. Gador, S. Ulrich, A. Stolow and T. J. Martı´ nez, J. Phys. Chem. A, 2007, 111, 8500. 23 C. E. Crespo-Herna´ndez, B. Cohen, P. M. Hare and B. Kohler, Chem. Rev., 2004, 104, 1977. 24 Y. He, C. Wu and W. Kong, J. Phys. Chem. A, 2004, 108, 943. 25 A. Yoshikawa and S. Matsika, Chem. Phys., 2008, 347, 393. 26 C. E. Crespo-Herna´ndez, B. Kohen and B. Kohler, Nature, 2005, 436, 1141. 27 N. Gador, E. Samoylova, V. R. Smith, A. Stolow, D. M. Rayner, W. Radloff, I. V. Hertel and T. Schultz, J. Phys. Chem. A, 2007, 111, 11743. 28 A. W. Lange and J. M. Herbert, J. Am. Chem. Soc., 2009, 131, 3913. 29 H. Fernando, G. A. Papadantonakis, N. S. Kim and P. R. LeBreton, Proc. Natl. Acad. Sci. U. S. A., 1998, 95, 5550. 30 P. Slavı´ cˇek, B. Winter, M. Faubel, S. E. Bradforth and P. Jungwirth, J. Am. Chem. Soc., 2009, 131, 6460. 31 C. Crespo-Herna´ndez, D. M. Close, L. Gorb and J. Leszczynski, J. Phys. Chem. B, 2007, 111, 5386. 32 A. A. Golubeva and A. I. Krylov, Phys. Chem. Chem. Phys., 2009, 11, 1303. 33 H. Sugiyama and I. Saito, J. Am. Chem. Soc., 1996, 118, 7063. 34 F. Prat, K. N. Houk and C. S. Foote, J. Am. Chem. Soc., 1998, 120, 845. 35 S. Schumm, M. Pre´vost, D. Garcia-Fresnadillo, O. Lentzen, C. Moucheron and A. Krisch-De Mesmaeker, J. Phys. Chem. B, 2002, 106, 2763. 36 A.-O. Colson, B. Besler and M. D. Sevilla, J. Phys. Chem., 1992, 96, 9787.

2306 | Phys. Chem. Chem. Phys., 2010, 12, 2292–2307

37 A.-O. Colson, B. Besler and M. D. Sevilla, J. Phys. Chem., 1993, 97, 13852. 38 D. Roca-Sanjua´n, M. Mercha´n and L. Serrano-Andre´s, Chem. Phys., 2008, 349, 188. 39 J. Berkowitz, J. Chem. Phys., 1978, 69, 3044. 40 P. Hobza, J. Sˇponer and T. Reschel, J. Comput. Chem., 1995, 16, 1315. 41 I. Dabkowska, P. Jurecˇka and P. Hobza, J. Chem. Phys., 2005, 122, 204322. 42 S. Grimme, J. Comput. Chem., 2004, 25, 1463. 43 S. Grimme, J. Comput. Chem., 2006, 27, 1787. 44 P. Jurecˇka and P. Hobza, J. Am. Chem. Soc., 2003, 125, 15608. 45 J.-D. Chai and M. Head-Gordon, Phys. Chem. Chem. Phys., 2008, 10, 6615. 46 W. C. Murray, N. C. Handy and G. J. Laming, Mol. Phys., 1993, 78, 997. 47 V. I. Lebedev, Zh. Vychisl. Mat. Mat. Fiz., 1975, 15, 48. 48 R. J. Bartlett, Int. J. Mol. Sci., 2002, 3, 579. 49 J. F. Stanton and J. Gauss, Adv. Chem. Phys., 2003, 125, 101. 50 A. I. Krylov, Annu. Rev. Phys. Chem., 2008, 59, 433. 51 S. Pal, M. Rittby, R. J. Bartlett, D. Sinha and D. Mukherjee, Chem. Phys. Lett., 1987, 137, 273. 52 J. F. Stanton and J. Gauss, J. Chem. Phys., 1994, 101, 8938. 53 M. Kamyia and S. Hirata, J. Chem. Phys., 2006, 125, 074111. 54 P. A. Pieniazek, S. A. Arnstein, S. E. Bradforth, A. I. Krylov and C. D. Sherrill, J. Chem. Phys., 2007, 127, 164110. 55 P. A. Pieniazek, S. E. Bradforth and A. I. Krylov, J. Chem. Phys., 2008, 129, 074104. 56 P. A. Pieniazek, A. I. Krylov and S. E. Bradforth, J. Chem. Phys., 2007, 127, 044317. 57 P. A. Pieniazek, J. VandeVondele, P. Jungwirth, A. I. Krylov and S. E. Bradforth, J. Phys. Chem. A, 2008, 112, 6159. 58 P. A. Pieniazek, E. J. Sundstrom, S. E. Bradforth and A. I. Krylov, J. Phys. Chem. A, 2009, 113, 4423. 59 A. Landau and A. I. Krylov, J. Chem. Phys., 2010, in press. 60 F. Weinhold and C. R. Landis, Chem. Ed.: Res. Pract. Eur., 2001, 2, 91. 61 A. A. Golubeva, P. A. Pieniazek and A. I. Krylov, J. Chem. Phys., 2009, 130, 124113. 62 Y. Shao, L. F. Molnar, Y. Jung, J. Kussmann, C. Ochsenfeld, S. Brown, A. T. B. Gilbert, L. V. Slipchenko, S. V. Levchenko, D. P. O’Neil, R. A. Distasio Jr, R. C. Lochan, T. Wang, G. J. O. Beran, N. A. Besley, J. M. Herbert, C. Y. Lin, T. Van Voorhis, S. H. Chien, A. Sodt, R. P. Steele, V. A. Rassolov, P. Maslen, P. P. Korambath, R. D. Adamson, B. Austin, J. Baker, E. F. C. Bird, H. Daschel, R. J. Doerksen, A. Drew, B. D. Dunietz, A. D. Dutoi, T. R. Furlani, S. R. Gwaltney, A. Heyden, S. Hirata, C.-P. Hsu, G. S. Kedziora, R. Z. Khalliulin, P. Klunziger, A. M. Lee, W. Z. Liang, I. Lotan, N. Nair, B. Peters, E. I. Proynov, P. A. Pieniazek, Y. M. Rhee, J. Ritchie, E. Rosta, C. D. Sherrill, A. C. Simmonett, J. E. Subotnik, H. L. Woodcock III, W. Zhang, A. T. Bell, A. K. Chakraborty, D. M. Chipman, F. J. Keil, A. Warshel, W. J. Herhe, H. F. Schaefer III, J. Kong, A. I. Krylov, P. M. W. Gill and M. Head-Gordon, Phys. Chem. Chem. Phys., 2006, 8, 3172. 63 K. Mu¨ller-Dethlefs and P. Hobza, Chem. Rev., 2000, 100, 143. 64 J. Sˇponer, J. Leszczynski and P. Hobza, Biopolymers, 2002, 61, 3. 65 H. Saigusa, J. Photochem. Photobiol., C, 2006, 7, 197. 66 M. S. de Vries and P. Hobza, Annu. Rev. Phys. Chem., 2007, 58, 585. 67 M. Kabela´cˇ and P. Hobza, Chem.–Eur. J., 2001, 7, 2067. 68 M. Kabela´cˇ and P. Hobza, J. Phys. Chem. B, 2001, 105, 5804. 69 J. Ran and P. Hobza, J. Phys. Chem. B, 2009, 113, 2933. 70 C. Plu¨tzer, I. Hu¨nig, K. Kleinermanns, E. Nir and M. S. de Vries, ChemPhysChem, 2003, 4, 838. 71 M. Kratochvil, O. Engkvist, J. Sponer, P. Jungwirth and P. Hobza, J. Phys. Chem. A, 1998, 102, 6921. 72 M. S. de Vries, in Radiation Induced Molecular Phenomena in Nucleic Acids, ed. M. Shukla and J. Leszczynski, Springer, Berlin, p. 323. 73 Chr. Plu¨tzer, E. Nir, M. S. de Vries and K. Kleinermanns, Phys. Chem. Chem. Phys., 2001, 3, 5466. 74 C. M. Oana and A. I. Krylov, J. Chem. Phys., 2007, 127, 234106. 75 H. Stoll, G. Wagenblast and H. Preub, Theor. Chim. Acta, 1980, 57, 169.

This journal is

 c

the Owner Societies 2010

76 K. B. Bravaya, O. Kostko, M. Ahmed, S. Dolgikh, A. Landau and A. I. Krylov, 2009; in preparation. 77 H. S. Park, S. H. Nam, J. K. Song, S. M. Park and S. Ryu, J. Phys. Chem. A, 2008, 112, 9023. 78 S. H. Nam, H. S. Park, S. Ryu, J. K. Song and S. M. Park, Chem. Phys. Lett., 2008, 450, 236. 79 L. P. Candeias and S. Steenken, J. Am. Chem. Soc., 1989, 111, 1094. 80 S. Steenken, Chem. Rev., 1989, 89, 503. 81 M. Hutter and T. Clark, J. Am. Chem. Soc., 1996, 118.

This journal is

 c

the Owner Societies 2010

82 A. K. Ghosh and G. B. Schuster, J. Am. Chem. Soc., 2006, 128, 4172. 83 A. Kumar and M. D. Sevilla, J. Phys. Chem. B, 2009; asap. 84 J. Bertran, A. Oliva, L. Rodrı´ guez-Santiago and M. Sodupe, J. Am. Chem. Soc., 1998, 120, 8159. 85 A. L. Sobolewski, W. Domcke and C. Ha¨ttig, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 17903. 86 S. Perun, A. L. Sobolewski and W. Domcke, J. Phys. Chem. A, 2006, 110, 9031. 87 I. Hu¨nig, C. Plu¨tzer, K. A. Seefeld, D. Lo¨wenich, M. Nispel and K. Kleinermanns, Comput. Phys. Commun., 2004, 5, 1427.

Phys. Chem. Chem. Phys., 2010, 12, 2292–2307 | 2307